The review is that of an Internet Society report by Michael Kende and Karen Rose, based on evidence from Rwanda. The objective of this study is to understand the impact of content hosting decisions (within the country vs. overseas), as well as to develop a practical guide on creating an attractive enabling environment for hosting content locally.
The paper defines and discusses the difference of locally relevant content and locally hosted content in Rwanda. Locally relevant content has proved to be aa factor that increases the overall use of the Internet in many economies studied by the Internet Society. The content however, needs to be accessible quickly (i.e. the page needs to load fast, therefore the lower the latency the better) and cheaply. As such, more attention has been given to creating locally relevant content. Yet, less attention has been given to the supporting infrastructure (e.g. transit prices / cost of International bandwidth). Content providers have no real obligation to host locally and lack the incentive for the same, as it is cheaper to host content overseas that in Rwanda. However, by hosting content internationally, access becomes costly for the end user (as internet service providers pass down the costs associated with traffic flowing back in to the country). The study also proves that hosting content overseas increases latency which in-turn impacts adoption and use as it deters the user experience.
The following table is data from one of the largest Rwandan websites. The cost saving by hosting overseas is USD 111 per annum. However, the cost of importing traffic which was approximately 9 Mbps, is USD 13,500 per annum. This deonstrates the large negative externality with hosting locally relevant content overseas. Further, the user experience is significantly affected as latency increases from 10 ms to 350 ms when content is hosted outside Rwanda.
Figure 1: The negative externality associated with overseas hosting of local content
Source: Promoting Local Content Hosting to Develop the Internet Ecosystem, Michael Kende and Karen Rose, 2015
The effects of Internet Exchange Points (IXPs) as seen previously in Kenya and Nigeria are then explored in the Rwandan context. Google Global Cache (GGC) was made available in Rwanda in mid-2013 and an Akamai cluster in 2014. User experiences were drastically improved and traffic through the Rwandan IXP increased by four-fold. After having analysed the evidence the authors of this study recommend hosting content locally in Rwanda. They also suggest conducting multi-stakeholder forums to raise awareness on the advantages of hosting content locally, among others.
However, the lack of (or perceived) quality of service for local hosting as well as level of security remain concerns for content owners to host locally.
To view the slide-set of this report review click here.
To view the report click here.